Indoor navigation using Wi-Fi fingerprinting combined with pedestrian dead reckoning

Abstract

This paper presents a method by which to calibrate the Wi-Fi fingerprinting database with less effort using smartphone based Pedestrian Dead Reckoning (PDR) without any a-priori knowledge. Because the accumulated errors from PDR will decrease the quality of the database, we employ a quaternion based orientation extended Kalman filter (EKF) to deal with the pedestrian heading and to narrow down the PDR positioning error to 2.2 meters for a 270 meter path. Furthermore, we implement the Walkie-Markie method to enhance the accuracy of the PDR step positions used to replace the reference points (RPs) built into the Wi-Fi fingerprinting database. The method defines the Wi-Fi landmarks according to the trend of the Receive Signal Strength (RSS) collected by PDR and converges the pathway map using the Wi-Fi Marks (WMs). In a simulation scenario, the WMs and PDR pathway errors is nearly 0 meters.